* Matt Brewer is a PhD candidate at Boston University. His research is in the philosophy of the geosciences, and his primary interests include geoscientific methodology, explanation, measurement, and theory pursuit. In his work, he focuses on scientific practices that make use of the theory of plate tectonics (e.g., geology, geomorphology, and geophysics). Matt’s dissertation is tentatively titled, “Evidence, Classification, and Explanation: Towards a Philosophy of Plate Tectonics.” He writes…
Even if you’re not a scientist (or, like me, a philosopher who thinks about science), you’ve likely heard the phrase “absence of evidence is not evidence of absence.” One place it shows up is in conversations about the existence of extraterrestrial life. The universe is so vast, the thinking goes, that even if we do not have evidence for the existence of aliens, this does not mean we should jump to the conclusion that aliens do not exist.
Indeed, the phrase is often attributed to cosmologists and astronomers. Consider this technical report from a 1971 NASA project on the feasibility of methods to detect extraterrestrial life. Its second chapter begins with an epigraph, “[a]bsence of evidence is not evidence of absence,” which the writers attribute to the famed British astronomer Martin Rees. Others attribute the phrase to Carl Sagan, who included it in a list of logical and rhetorical fallacies in his popular 1995 book, The Demon-Haunted World. Sagan characterizes arguments from absence as an “appeal to ignorance,” arising from “impatience with ambiguity.” While we lack evidence for other forms of advanced life, we should not conclude that we are therefore “central to the universe.”
To most, it is probably unsurprising that Sagan characterizes arguments from absence as fallacious. There are many cases in the history of science which show that appealing to the absence of evidence can lead to false conclusions. Consider “Lazarus taxa” like the coelacanth. Coelacanths were once thought to have gone extinct in the Cretaceous period due to their absence from the fossil record and the lack of living specimens, until a live coelacanth was caught off the coast of South Africa in 1938 (Smith 1939). So absence of evidence for living coelacanths was not reliable evidence of their absence.
Still, scientists in certain contexts sometimes do treat absence of evidence as evidence of absence. In this essay, I’d like to offer a novel defense of that practice, especially in the context of the paleosciences.* Before doing this, however, I’d like to further explore the history of the phrase, “absence of evidence is not evidence of absence.” The reason is that the phrase did not originate in the science of astronomy, after all. It came from geology.
[* This essay is adapted from a paper of mine, “When Should Absence of Evidence Be Evidence of Absence? A Case Study from Paleogeology.” It is forthcoming in Philosophy of Science.]
* * *
The phrase “absence of evidence is not evidence of absence” was coined in the waning years of the nineteenth century. While close approximations can be found in 1887 and 1891 (see here for a detailed history), the complete expression first appeared in an 1895 issue of The Glacialists’ Magazine. At the time, glaciology was emerging as a distinct subfield of geology. The first glaciologists were fascinated by “erratics,” or large boulders thought to have been deposited by moving glaciers (Schroder 2023). A major challenge was to explain how glaciers could have transported these rocks over vast distances, and sometimes up mountain slopes.
Some geologists, such as William Sollas, devised clever means to simulate glaciers in the laboratory. Sollas used wax to model the viscosity of ice, and his experimental results demonstrated the possibility that moving ice could transport rocks from lower to higher elevations. Skeptics of glacial deposition had previously cited the lack of a plausible mechanism as evidence against the hypothesis. However, Sollas dismissed the skeptics when he published his results in a February issue of The Quarterly Journal of the Geological Society of London. In a pithy rejoinder, Sollas refuted those who would, according to him, “treat absence of evidence as evidence of absence” (Sollas 1895).
Still, the precise phrase “absence of evidence is not evidence of absence” did not occur until the geologist Thomas Sheppard’s article in the December issue of The Glacialists’ Magazine (Sheppard 1895). Sheppard was writing in response to Henry Howorth, who had published his own thoughts on glacial erratics the previous year (Howorth 1894). A skeptic of the glacial hypothesis, Howorth argued that erratic stones were instead dropped by Danish pirates who had used them as anchors for their ships. In support of this view, Howorth cited the absence of a particular line of evidence for glaciers: “terminal moraines,” or large collections of sediment that accrue at the front end of a moving glacier, much like a bulldozer. In reply to Howorth, Sheppard warned that “[e]ven if no moraine existed, it does not follow that there was no Ice-sheet.” He pointed out that “it has been remarked by Mr. Dugald Bell that ‘absence of evidence is not evidence of absence.’”
Sheppard’s use of the expression is its first known occurrence. While he attributed it to the Scottish geologist Dugald Bell, Sheppard did not provide a precise citation. Thus far, I have not found any instance of Bell using the phrase in print (but if you come across it, please let me know!). Bell’s biographers in a 1908 history of the Geological Society of Glasgow remark that “[h]is large acquaintance with books and his retentive memory served him well, however, so that when occasion required he could quote from many an author, or cap the lines or verses given by others” (Macnair and Mort 1908). Perhaps Bell had read the phrase in some forgotten source and began to use it himself. Still, we can look to his published work to see how he treated arguments from absence in his own geological practice.
In an 1897 article published in Geological Magazine, Bell considered another debate that was relevant to glacial deposition. Scottish geologists of the time disagreed in their interpretations of what Bell called “shelly clays,” or beds of sediment containing the shelly remains of marine life. Some geologists interpreted the shelly sediments as evidence that Scotland had undergone a “great submergence” underwater at some point in its past. Others argued that the land had not been submerged, and that the shelly remains were picked up from elsewhere and deposited by glaciers as they crawled across the land. In a series of writings published tbetween 1870 and 1900, Bell argued against the “great submergence” theory. It seems that Bell dedicated much of his energy to this cause: his biographers remarked that Bell was “always ready to do battle for his pet theory, the cause of land ice against that of submergence” (Macnair and Mort 1908).
Now an irony. To dismiss the theory of a “great submergence,” Bell appealed to absence of evidence as evidence of absence. In his writings, he noted the various traces of evidence that would be expected had submergence occurred, such as “wave-worn cliffs, well-marked beaches,” and fossil evidence of other marine life. The absence of evidence was “conspicuous” in this case, because “if the sea had reached the level supposed, […] it should have left traces of itself in innumerable localities all over the country” (Bell 1897).
It seems that glaciologists in the late nineteenth century had a nuanced understanding of arguments from absence. At least Dugald Bell understood that absence of evidence is in fact evidence of absence in cases when there is a high expectation of finding evidence. When scientists have good reasons to believe that they would find evidence had the phenomenon in question occurred, absence of evidence can be strong evidence of absence.
* * *
In recent work, I have related this nineteenth-century conception to the views of some contemporary philosophers of science. According to Elliott Sober, for example, absence of evidence provides evidence of absence whenever it is highly likely that evidence of a certain type will be found (Sober 2009). By contrast, when finding evidence is unlikely, absence of evidence is so expected that its evidential value is negligible. Similarly, Efraim Wallach argues that in archaeological contexts, absence of evidence can be compelling as evidence of absence, because archaeologists have good reasons to believe that traces of past settlements are likely to endure (Wallach 2019). Because the evidential strength of absence of evidence depends upon the probability of finding evidence, I call this conception the “Probabilistic View.”
The Probabilistic View is sensible. Still, I argue that it does not capture some important considerations in the paleosciences. The view falls short because in paleoscientific contexts, finding evidence is often highly unlikely; nevertheless, arguments from absence are sometimes highly fruitful. I argue that appeals to negative evidence are warranted in these cases because they offer theoretical scaffolds to investigate auxiliary hypotheses. I illustrate this with a case study from paleogeology concerning the timing of the onset of plate tectonics.
There is no agreed upon answer to the question, “when did plate tectonics begin on Earth?” Scientists have posited dates that span billions of years, from the Neoproterozoic era (ca. 538 million to 1 billion years ago) (e.g., Hamilton 2011) to the Hadean era (ca. 4 billion years ago) (e.g., Hopkins et al. 2009). Answers diverge so widely because scientists differ in how they interpret the absence of key lines of evidence for plate tectonics. Some kinds of rock, such as blueschist and ophiolite, are interpreted as key evidence for the operation of plate tectonics at a given time in Earth’s history. Many scientists think that these rocks form only under conditions that are characteristic of plate tectonics. Also, they are absent from the geologic record prior to the Neoproterozoic era.
Some geologists like Robert Stern have cited the absence of blueschist and ophiolite as evidence that plate tectonics did not begin until the Neoproterozoic. In other words, these geologists interpret the absence of evidence for plate activity as evidence of absence. Others are more skeptical. They invoke the phrase “absence of evidence is not evidence of absence” and caution that blueschist and ophiolite might have existed prior to the Neoproterozoic but not been preserved (e.g., Palin et al. 2020). Due to the uncertain nature of geologic preservation, critics of Stern advise against treating absence of evidence as evidence of absence.
While Stern’s critics are right to emphasize that evidence from Earth’s deep past cannot be highly expected due to the vicissitudes of the rock record, I think that Stern is justified in making an argument from absence here. Even if the hypothesis is likely to be false, it may still be useful to advance the hypothesis insofar as it generates other epistemic goods. The Probabilistic View is blind to such considerations, and because of this is prone to discourage the pursuit of fruitful hypotheses. That is: if absence of evidence is ignored as evidence of absence, then productive hypotheses like Stern’s may be unlikely to gain a toehold.
To show why the pursuit of a hypothesis like Stern’s is fruitful despite the admonitions of the Probabilistic View, it is useful to highlight two auxiliary assumptions that Stern makes to support his argument from absence.
To argue that plate tectonics initiated during the Neoproterozoic, Stern must address what he calls “preservation bias” (Stern 2018). He must show that the absence of evidence is not due to poor preservation, lest we think that plate tectonics initiated earlier without leaving any extant traces. Indeed, Stern must make the case that the evidence is genuinely absent (i.e., it was never present). One cause of poor preservation is erosion: after rocks such as blueschist are formed deep underground, they are exhumed, and in some cases, they are completely destroyed by erosion. However, Stern relies on a key auxiliary assumption: blueschist and ophiolite are affected differently by erosion. If erosion was so impactful that evidence was destroyed, then we would not expect to see correlations in the frequencies of blueschist and ophiolite across time, because each line of evidence would be destroyed to a different extent by erosion. Nevertheless, Stern points out that we do observe correlations in the frequencies of blueschist and ophiolite across time. Given such correlations, Stern concludes that “preservation bias” is not a concern for these key lines of evidence. Following his appeal to absence of evidence, Stern’s auxiliary assumption (that erosion affects evidence differently) stands as a potential target of inquiry by future investigators.
In a subsequent argument (2023), Stern relies on another auxiliary assumption related to “preservation bias.” He argues that if a given period in Earth’s history is bookended by two periods of adequate preservation, then we can assume that the intervening period also adequately preserved evidence. Any absence of evidence in the intervening period between periods of adequate preservation is thus a genuine absence. In addition to clear evidence for the operation of plate tectonics during the Neoproterozoic, Stern amends his earlier position and argues that there is also clear evidence for plate tectonics during the earlier Orosirian period (during the Paleoproterozoic). Because evidence is adequately preserved for the Neoproterozoic and the Orosirian, Stern assumes that evidence should also be adequately preserved during the interim. However, we lack evidence for plate tectonics during the intervening time. Thus, Stern concludes that the Earth has experienced multiple episodes of plate tectonics, punctuated by a “boring” period of little tectonic activity in the interim. Stern’s auxiliary assumption about bookended preservation will likely be investigated by other geoscientists in response to his appeal to absence of evidence.
Perhaps some readers already have spotted a potential objection to Stern’s assumptions. You might be thinking that he is too uniformitarian about the preservation of evidence. For instance, many scientists have given alternative arguments to explain the lack of preservation of rocks such as blueschist. Others have argued that the conditions of the early Earth were so different that plate tectonic processes did not form blueschist (e.g., Palin and White 2015). Furthermore, perhaps a catastrophic event could destroy evidence between two periods of adequate preservation. My worry is that without in the first place appealing to absence of evidence as evidence of absence, such alternative explanations might never be pursued, leaving scientists deprived of a richer understanding of their target phenomena. For example, in this case, scientists gain insights into the metamorphic processes that form blueschist by seeking to explain its absence. If the absence of blueschist were not considered, then such insights might remain unknown.
* * *
In my forthcoming paper, I defend what I call a Pragmatic View of arguments from absence. The basic idea is that appeals to absence of evidence as evidence of absence are warranted in the paleosciences, because they offer a theoretical scaffold to investigate auxiliary hypotheses. Auxiliary hypotheses are conditionals: “If H [our initial hypothesis] is true, then evidence y will be observed.” To test auxiliary hypotheses, scientists must entertain an initial hypothesis over other conceivable alternatives. Entertaining absence of evidence as evidence of absence allows Stern to consider auxiliary assumptions related to the preservation of evidence. Others have attempted to refute him by specifying a wider range of alternatives with their own auxiliaries, against which Stern’s hypothesis might be contrasted. Treating absence of evidence as evidence of absence provides the initial reason (in some cases, perhaps the only reason) for investigating the implications of some hypotheses that might otherwise be neglected. If geologists strictly adhered to the Probabilistic View, then Stern might never have offered his arguments, and geologists would be deprived of productive lines of inquiry that have arisen from his appeals.
Ultimately, I do not wish to dismiss the Probabilistic View. I simply wish to point out that in many investigations, absence of evidence guides scientists by focusing their investigations on possible alternatives and auxiliaries which might explain the absence. If scientists did not treat absence of evidence as evidence of absence, then there would be comparatively little warrant for these investigations. Yet, these investigations allow scientists to make genuine progress towards more refined models for the Earth’s deep past.
The debates that preoccupy today’s paleogeologists are reminiscent of the debates that were the focus of the pioneers of glaciology in the late nineteenth century. In both contexts, much effort is dedicated to explaining the absence of certain lines of trace evidence. Doing so is not a trivial matter: just as plate tectonics is thought to destroy evidence as it “recycles” the Earth’s crust, some glaciologists posited that moving glaciers could destroy evidence for past episodes of submergence as they crawled across the land (Bell 1891). Glaciologists understood that absence of evidence is evidence of absence when finding evidence is highly expected. Still, even when the probability of finding evidence is low or unclear, I think that appeals to absence of evidence as evidence of absence have a place in paleogeological thinking. Such appeals give rise to fruitful pursuits of possible explanations for the absence in question.*
[* Matt will present his paper on the Pragmatic View at the meeting of the Philosophy of Science Association in November. He intends to develop an expanded version of the paper as a chapter in his dissertation. Plate tectonics is a complex, global phenomenon that has operated on Earth for millions (perhaps billions) of years, and geoscientific practices that make use of the theory are ripe for philosophical analysis. Stay tuned for more work on the “philosophy of plate tectonics!” And watch out for our next post on Extinct, which will continue the discussion of negative evidence in the historical sciences.]
References
Bell, D. 1891. Phenomena of the Glacial Epoch: II. The 'Great Submergence.' Transactions of the Geological Society of Glasgow 9, no. 1: 100-38. https://doi.org/10.1144/transglas.9.1.100.
Bell, D. 1897. The 'Great Submergence' Again: Clava. Geological Magazine 4, no. 1: 27-30. https://doi.org/10.1017/S001675680017503X.
Hamilton, W. B. 2011. Plate tectonics began in Neoproterozoic time, and plumes from deep mantle have never operated. Lithos 123, nos. 1-4: 1-20. https://doi.org/10.1016/j.lithos.2010.12.007.
Hopkins, M. T., Harrison, M. and Manning, E. C. 2008. Low heat flow inferred from >4 Gyr zircons suggests Hadean plate boundary interactions. Nature 456: 493-96. https://doi.org/10.1038/nature07465.
Howorth, H. H. 1894. The North Sea Ice Sheet. Nature 50, no. 1282: 79. https://doi.org/10.1038/050079a0.
Jamieson, T. F. 1874. On the Last Stage of the Glacial Period in North Britain. Quarterly Journal of the Geological Society of London 30, nos. 1-4: 317-38. https://doi.org/10.1144/GSL.JGS.1874.030.01-04.40.
Macnair, P., and Mort, F. eds. 1908. History of the Geological Society of Glasgow, 1858-1908, with Biographical Notes of Prominent Members. Glasgow: Geological Society of Glasgow.
NASA (1972). Project Cyclops: a Design Study of a System for Detecting Extraterrestrial Intelligent Life. https://ntrs.nasa.gov/citations/19730010095.
O’Toole, G. 2019. Absence of Evidence Is Not Evidence of Absence. Quote Investigator. https://quoteinvestigator.com/2019/09/17/absence/.
Palin, R. M., and White, R. W. 2015. Emergence of blueschists on Earth linked to secular changes in oceanic crust composition. Nature Geoscience 9: 60-64. https://doi.org/10.1038/ngeo2605.
Palin, R. M., Santosh, M., Cao, W., Li, S., Hernández-Uribe, D. and Parsons, A. 2020. Secular change and the onset of plate tectonics on Earth. Earth-Science Reviews 207, 103172: 1-41. https://doi.org/10.1016/j.earscirev.2020.103172.
Sagan, C. and Druyan, A. 1995. The Demon-Haunted World: Science as a Candle in the Dark. New York: Ballantine Books.
Schroder, J. 2023. The Remarkable Reverend Henry W. Crosskey, the Erratic Blocks Committee, and the Crosskey Collection of Glacial Erratic Specimens. Birmingham Erratic Boulders Project. https://erraticsproject.org/henry_crosskey/.
Sheppard, T. 1895. On the Occurrence of Scandinavian Boulders in England. The Glacialists' Magazine 3, no. 1: 129-32. https://babel.hathitrust.org/cgi/pt?id=uc1.b4182734&seq=149.
Smith, J. L. B. 1939. A Living Fish of Mesozoic Type. Nature 143: 455-56. https://doi.org/10.1038/143455a0.
Sober, E. 2009. Absence of evidence and evidence of absence: evidential transitivity in connection with fossils, fishing, fine-tuning, and firing squads. Philosophical Studies 143: 63-90. https://doi.org/10.1007/s11098-008-9315-0.
Sollas, W. J. 1895. An Experiment to Illustrate the Mode of Flow of a Viscous Fluid. Quarterly Journal of the Geological Society of London 51: 361-68. https://doi.org/10.1144/GSL.JGS.1895.051.01-04.28.
Stern, R. J. 2018. The evolution of plate tectonics. Philosophical Transactions of the Royal Society A 376, no. 20170406: 1-21. https://doi.org/10.1098/rsta.2017.0406.
Stern, R. J. 2023. The Orosirian (1800–2050 Ma) plate tectonic episode: Key for reconstructing the Proterozoic tectonic record. Geoscience Frontiers 14, no. 3: 101553. https://doi.org/10.1016/j.gsf.2023.101553.
Wallach, E. 2019. Inference from absence: the case of archaeology. Palgrave Communications 5, no. 94: 1-10. https://doi.org/10.1057/s41599-019-0307-9.